Software engineer (TimescaleDB) posting about AI coding agents, enforceable/testable architecture (e.g., ArchUnit), and quality‑driven engineering practices.
Claude seems to take forever planning today. 15 minutes+ and then its like: I need to set height:500px (which wasn't even correct it needed max height). I should probably not have used plan mode for that but usually it was smart enough to realize it was a tiny change?
I still never hit the Max subscription token limit. Do people just never /clear their sessions and always operate at 1million input tokens? I barely ever go beyond 30% context window and the model is already dumb by then

Forcing Claude to be structured about it's process and it's like: "THIS IS SO TRIVIAL THOUGH". Truly how I would react, but truth is it would miss stuff otherwise.
The highest-performing engineers I know never say "that's not my job." They ship the feature. Design it. Test it. Deploy it. Tech spent years splitting one feature across 6 people. AI is collapsing that back down and the people who embrace it...
Velocity is a vector of speed and direction. If you use LLMs only to improve speed without carefully aligning the direction you'll end up building something useless.
I love that clanker for LLMs and coding agents is catching on. I always imagine some old rusty robot hacking away on one of those beige gray keyboards in the darkness now.
Clickhouse is trying to push postgres + clickhouse as the ultimate analytics DB stack. But tbh adding an eventually consistent database to your stack that you needed to sync too is anything but trivial. Love the product but I'd just use...
Thinking about setting up automatic demo deployments for PRs on my side project Kviklet. I do review all code that Claude Code produces and also click through the main flows manually locally. But it's always a fetch branch, restart stuff, potentially...
In 2023 I mostly used Copilot and ChatGPT to code In 2024 Sonnet 3.5 came out and became my daily driver. I tried a few things like Windsurf and Cursor but nothing had as big of an impact as the model...
Interesting additional thought about this: I use archunit https://www.archunit.org/ to force Claude to follow some patterns (never access the DB from the service layer for example, never return database package entities from the controller). I wonder if we should have more...